cases and match-condition (#49)

Author: geritwagner

bib_dedupe/match_conditions.py

@@ -110,4 +110,6 @@ def both_entrytypes(entrytype: str) -> str:
     f' ~({CONTAINER_TITLE}_2.str.contains("conf") | {CONTAINER_TITLE}_2.str.contains("work") | {CONTAINER_TITLE}_2.str.contains("proc")) ) & '
     f' ( ({CONTAINER_TITLE}_1.str.contains("conf") | {CONTAINER_TITLE}_1.str.contains("work") | {CONTAINER_TITLE}_1.str.contains("proc")) & '
     f'  ~{CONTAINER_TITLE}_1.str.contains("j") ))',
+    # Inproceedings: more sensitive to year mismatches
+    f'({both_entrytypes("inproceedings")} & {mismatch(YEAR)})',
 ]

tests/test_cases.json

@@ -439,6 +439,94 @@
         "booktitle": "Americas Conference on Information Systems"
       },
       "expected_duplicate": false
+    },
+    {
+      "id": "deng_taylor_moussawi_hicss_2021_vs_2022_minitrack_intro",
+      "note": "Annual HICSS minitrack introductions share title/authors but are different years and different DOIs -> should NOT match.",
+      "record_a": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "id_0000003",
+        "colrev_status": "md_processed",
+        "author": "Deng, Xuefei and Taylor, Joseph and Moussawi, Sara",
+        "title": "Introduction to the Minitrack on Crowdsourcing and Digital Workforce in the Gig Economy",
+        "year": "2021",
+        "booktitle": "Proceedings of the Annual Hawaii International Conference on System Sciences",
+        "doi": "10.24251/HICSS.2021.510"
+      },
+      "record_b": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "id_0000004",
+        "colrev_status": "md_processed",
+        "author": "Deng, Xuefei and Taylor, Joseph and Moussawi, Sara",
+        "title": "Introduction to the Minitrack on Crowdsourcing and Digital Workforce in the Gig Economy",
+        "year": "2022",
+        "booktitle": "Proceedings of the Annual Hawaii International Conference on System Sciences",
+        "doi": "10.24251/HICSS.2022.544"
+      },
+      "expected_duplicate": false
+    },
+    {
+      "id": "erroneous_conference_merge_hall_dianne_2020_unknown_vs_hall_gillian_1995",
+      "note": "Same venue, but different authors/years (2020 vs 1995); record_a has UNKNOWN title -> must NOT match.",
+      "record_a": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "0000006472",
+        "author": "Hall, Dianne and Lee, Kangbok and Han, Sumin and In, Joonhwan",
+        "year": "2020",
+        "title": "UNKNOWN",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "record_b": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "0000005291",
+        "author": "Hall, Gillian",
+        "year": "1995",
+        "title": "Negotiation in Database Schema Integration",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "expected_duplicate": false
+    },
+    {
+      "id": "reynoso_2010_unknown_title_vs_reynoso_2013_spanish_title",
+      "note": "Similar author strings but different years and (one) unknown title vs specific title -> must NOT match.",
+      "record_a": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "0000005436",
+        "author": "Reynoso, Gómez and Manuel, Juan and Lizbeth, Estela and Andrade, Muñoz and Macías Díaz, Jorge Eduardo",
+        "year": "2010",
+        "title": "UNKNOWN",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "record_b": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "0000009531",
+        "author": "Reynoso, Gómez and Manuel, Juan and Calzada De Luna, Américo C and Lizbeth, Estela and Andrade, Muñoz",
+        "year": "2013",
+        "title": "Medición Experimental del Desempeño de las Aplicaciones Bajo Ambientes de Red",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "expected_duplicate": false
+    },
+    {
+      "id": "erroneous_conference_merge_patel_1999_vs_2025",
+      "note": "Clearly different authors/years (1999 vs 2025) but both AMCIS inproceedings -> must NOT match.",
+      "record_a": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "0000008147",
+        "author": "Patel, Nandish V",
+        "year": "1999",
+        "title": "Developing Tailorable Information Systems through Deferred System's Design",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "record_b": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "011979",
+        "author": "Patel, Nadya Shaznay and Avnit, Karin and Koh, Jeffrey T K V and Lim, Jawn and Tay, Peter and Supian, Hedirman and Kwan, Jeffrey Tzu and Koh, Valino and Chai, Kay",
+        "year": "2025",
+        "title": "UNKNOWN",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "expected_duplicate": false
     }
   ]
 }